Data collected from volunteers' own smartphones has the potential to emulate randomized trials. Interpreting this data can boost research into human behavior, revealed a new study.
Fani Tsapeli from the University of Birmingham, and her colleague and Mirco Musolesi from the University College London used user-generated data, harvested from their phones to evaluate the cause of increased stress levels of participants.
Most of the earlier research works relying on smartphones focused on detecting factors in the features extracted from smartphone data. But that pure correlation analysis did not provide for a sufficient understanding of human behavior.
In this study, the authors used data from a research project at Dartmouth College, Hanover, US, called StudentLife. It included information on participants' location taken from raw GPS data, which helped determine whether they were working or socializing.
Also included was data on activity levels, like running, walking or traveling on public transport, inferred from participants' raw accelerometer data.
They found that exercising and spending time outside the home and working environment have a positive effect on participants' stress levels.
By contrast, they found that reduced working hours only slightly impact stress. The conclusions cannot be extended to the general population due to the small sample size. But the approach has been validated and shows great promise for further studies.
The study was published recently in EPJ Data Science.